IRC channel logs

<damo22>coreboot sets up the intel gpu as a framebuffer and uses seabios which has SeaVGABIOS : that exposes the existing framebuffer as a VESA device

<damo22>fbdev wont work

<damo22>but now i have reused the grub framebuffer in multiboot table to pass through to hurd console

<almuhs>great

<almuhs>i have researched about opengl bug. I found that, some years ago, there was some games which showed the same error than we have now in wesnoth and others

<damo22>so the steps are coreboot -> gfxinit -> seavga VESA -> grub -> X or hurd console

<damo22>but X gets mode from VESA

<almuhs>maybe it's necessary to compile the games with a specific flag?

<almuhs>ok

<damo22>opengl mesa has swrast

<almuhs>what is swrast?

<damo22>software rasteriser driver for opengl

<almuhs>yes

<damo22>there may be a way to force that mode

<damo22>using environment variable

<almuhs>but, why it works with some games but not with others?

<damo22>missing env var

<almuhs>i tried to forced OpenGL indirect, but then the DE doesn't boots

<damo22>you can force it inside X terminal

<damo22>not to start x

<almuhs>ok, i added that to xorg.conf

<damo22>try to get output from glxinfo

<almuhs>like that

<almuhs> https://askubuntu.com/a/932912

<damo22>LIBGL_DEBUG=verbose glxinfo

<almuhs>ok, i'm installing glxinfo

<damo22>mesa-utils

<almuhs>yes

<almuhs>but i have the same error

<damo22>what error

<almuhs>pasteboard doesn't works, so I'm searching how to see the image

<almuhs>Xlib: sequence lost ... and a failure in X_GlxIsDirect

<damo22>apt install pastebinit

<damo22>then you can pipe commands into | pastebinit to paste it to paste.debian

<damo22>wow my console works! startx and then kill the xserver returns to graphical console!

<almuhs> https://paste.debian.net/hidden/2b1039f2

<almuhs>shit

<almuhs> https://paste.debian.net/hidden/04ccb5a9

<almuhs>this

<almuhs>i had to remember how to redirect error output using a pipe

<almuhs>this is the glxinfo output

<almuhs>with verbose options I get the same

<damo22>ok so you need to enable indirect glx

<almuhs>i tried to enable in xorg.conf, but then the DE didn't start

<almuhs>i put that

<almuhs>Section "ServerFlags" Option "AllowIndirectGLX" "on" Option "IndirectGLX" "on" EndSection

<almuhs>i don't know how to set in the console

<damo22>Xorg -iglx ?

<damo22>i have the same problem

<damo22>i think the correct flag is +iglx

<damo22>but it doesnt work here

<almuhs>but, where can i apply this flag?

<almuhs>because i need to run a DE

<damo22>instead of startx

<damo22>Xorg +iglx +extension GLX

<damo22>(II) IGLX: Loaded and initialized swrast

<damo22>(II) GLX: Initialized DRISWRAST GL provider for screen 0

<damo22>but the DE doesnt start

<almuhs>in anywhere i have to indicate the DE

<almuhs>or i have to set a value in xorg.conf

<almuhs>now Xorg is broken. Even don't boot with startx

<almuhs>this afternoon i got the same, but i don't remember how i solved it

<damo22>its not broken, it is on but no drawing is occurring

<damo22>glxgears reports working

<almuhs>the DE don't boot

<damo22>$ glxgears -display :0

<damo22>21103 frames in 5.1 seconds = 4145.972 FPS

<damo22>16280 frames in 5.0 seconds = 3256.000 FPS

<damo22>but it doesnt appear on the screen

<damo22>its black

<almuhs>yeah, and no mouse

<almuhs>or mouse but no move

<almuhs>so, where can be the problem now?

<almuhs>now the games doesn't show glx error, but due to i can only run the command from ssh, i can't see in the screen

<almuhs>pruebas@debian-hurd:~$ glxinfo -display :0

<almuhs>name of display: :0

<almuhs>Xlib: sequence lost (0x10014 > 0x16) in reply type 0x0!

<almuhs>Xlib: sequence lost (0x10019 > 0x1b) in reply type 0x0!

<almuhs>Xlib: sequence lost (0x1001e > 0x20) in reply type 0x0!

<almuhs>Xlib: sequence lost (0x10023 > 0x25) in reply type 0x0!

<almuhs>Xlib: sequence lost (0x10028 > 0x2a) in reply type 0x0!

<almuhs>display: :0 screen: 0

<almuhs>direct rendering: No (If you want to find out why, try setting LIBGL_DEBUG=verbose)

<almuhs>server glx vendor string: SGI

<almuhs>server glx version string: 1.4

<almuhs>Xorg is fixed, but we need to fix the DE configuration to support it

<damo22>yeah so the game is working but its not displaying anything...

<almuhs>yeah. because i can get a menu or terminal in the desktop

<almuhs>can'

<almuhs>can't

<almuhs>i don't know if, from ssh, i can redirect the image to the display

<damo22>xinitrc

<damo22>has a startup script for Xorg

<almuhs>yes, i use that to start openbox

<almuhs>with exec

<almuhs>exec openbox-session

<almuhs>but, if i enable indirect, openbox doesn't load

<almuhs>neither any other desktop

<damo22>same

<almuhs>and Xorg bug?

<almuhs>is this a Xorg bug?

<almuhs>maybe lacks any irq

<damo22>idk

<damo22>it shouldnt need anything to write to the framebuffer, just its address and w h

<almuhs>we need to debug Xorg

<almuhs>but not today: 6:35 AM in Spain, i have to go sleep

<almuhs>bye

<damo22>#0 __strlen_sse2_bsf () at ../sysdeps/i386/i686/multiarch/strlen-sse2-bsf.S:49

<damo22>#1 0x01be6d9a in ?? () from /lib/i386-gnu/libxkbcommon.so.0

<damo22>#2 0x01be8c1f in xkb_keymap_mod_get_index () from /lib/i386-gnu/libxkbcommon.so.0

<damo22>#3 0x01bea9d2 in xkb_state_mod_names_are_active () from /lib/i386-gnu/libxkbcommon.so.0

<damo22>#4 0x01bc1c3b in execute_action (keycode=107) at ../../console-client/xkb/xkb.c:118

<damo22>when i press delete key i get segfault

<damo22>but backspace works

<damo22>youpi: i think i found a bug in libxkbcommon0 state.c:

<damo22> va_start(ap, match);

<damo22> while (1) {

<damo22> xkb_mod_index_t idx;

<damo22> const char *str = va_arg(ap, const char *);

<damo22> if (str == NULL)

<damo22> break;

<damo22> idx = xkb_keymap_mod_get_index(state->keymap, str);

<damo22>somehow str == 0x0, but it does not break, it continues to call xkb_keymap_mod_get_index with a NULL str

<damo22>must be overrun stack?

<damo22>buffer

<damo22>process_keypress_event (keycode_t keycode)

<damo22>char buf[100];

<damo22> https://paste.debian.net/hidden/018c6fff/

<azert>damo22: did you fix the 80x25 problem? On vga I think it is hardcoded define VGA_DISP_WIDTH, that should be a variable

<azert>Also, do you display the mouse pointer?

<damo22>yes mouse pointer works

<damo22>i did make the VGA_DISP_WIDTH detected

<damo22>but something else is restricting the width and height

<damo22>perhaps it uses ncursesw

<damo22>azert: is it legitimate to pass "blah" as a vararg?

<damo22>how does that work?

<azert>I don’t understand the question

<azert>vararg is syntactic sugar for passing as many parameters as one wants

<damo22>xkb_state_mod_names_are_active(struct xkb_state *state, enum xkb_state_component type, enum xkb_state_match match, ...)

<damo22>the 4th argument is a const string

<azert>Do you mean is it legitimate to pass a string literal as a vararg?

<damo22>yes

<azert>That would be passing a pointer

<azert>I think it is

<damo22>its strange because the pointer is NULL

<damo22>but its not detecting the null

<damo22>gdb says its NULL

<damo22> const char *str = va_arg(ap, const char *);

<damo22> if (str == NULL)

<damo22> break;

<damo22> idx = xkb_keymap_mod_get_index(state->keymap, str);

<damo22>its supposed to break

<azert>It’s probably gdb that is confused

<damo22>i dont think it is, its definitely calling the xkb function with a null pointer

<azert>Can you check the stack directly?

<damo22>yes

<damo22>actually, how do i inspect the 4th arg

<damo22>in the gdb frame

<damo22>#3 0x01bea9d2 in xkb_state_mod_names_are_active (state=0x2003af10, type=XKB_STATE_MODS_EFFECTIVE, match=XKB_STATE_MATCH_ALL)

<azert>x86 passes them all on the stack

<damo22>0x0000006b

<damo22>its not NULL!

<damo22>its the scan code

<azert>Ok

<damo22>but its supposed to be a pointer to a const string

<azert>It’s treated as an array

<azert>I think you found a bug

<azert>When you declare a literal string on the stack, it may be not a pointer

<azert>Try adding an explicit char* = string in the caller

<azert>So you pass the actual pointer

<damo22>ok i fixed it kindof

<damo22>i worked hard on this to get it over the line and mailed in, i actually go back to work tomorrow, ive been on holiday

<damo22>goodnight

IRC channel logs

2024-10-27.log